Kay Schluehr wrote: > You'll most likely need a GLR parser. i'm not sure why you think this. as far as i can tell, the OP needs a parser that is suitable for whatever grammar they find (and the grammar will probably be written for a particular parser, which may not be GLR).
however, if you are saying that only a GLR parser can parse natural languages then i think you are wrong. not only can grammars be rewritten in different ways, but a recursive descent parser with appropriate memoisation is capable of parsing "any" grammar. see, for example, the second example at http://www.acooke.org/lepl/advanced.html#memoisation - that is a left-recursive, highly ambiguous grammar, and is parsed successfully with recursive descent (as far as i can tell). for more info see http://www.acooke.org/lepl/implementation.html#memoisation and http://www.cs.uwindsor.ca/~hafiz/p46-frost.pdf in theory (if not currently in practice for my code, at least) it is also efficient (in a "big O" sense). disclaimer - the lepl parser linked to is my own and that functionality is very new (there's a beta version released, but it is buggy). however, that doesn't mean this is not possible.... andrew -- http://mail.python.org/mailman/listinfo/python-list
